Aim. The practice of dependability calculation and analysis occasionally deals with technical systems for which the dependability model is difficult or impossible to adequately describe with a set of serial and parallel connections and corresponding mathematical tools of multiplication of probabilities. The article examines the method of modeling the dependability of highly integrated systems through the analysis of the position of time to failure density function centroid f(t). This work is the continuation of a big research dedicated to the analysis of the properties of a density function centroid in highly integrated technical systems. In the first part it was shown that the centroid allows identifying the level of mutual influence of subsystems of a mechatronic system and identifying their contribution into the overall level of dependability of a whole product, where the primary criterion is the proximity of the partial centroid of the density function of each subsystem to the overall average centroid of the whole system. This paper assumes that the average centroid for a composition of density functions of product components does not depend on the way they are connected in the dependability model and thus can be used as the conditional reliability indicator for systems with fuzzy structural and functional connections. Methods. The research is based on graphs of time to failure density functions for conditional components of a complex technical system, such as electronics, mechanics and software. The diverse nature of the system’s components is reflected through the variation of parameters of the Weibull-Gnedenko law. In order to simplify the calculation and presentation of the results, the analysis is conducted not in an integrated manner for 3 components, but for pairs. For each pair of subsystems density functions are calculated and plotted both for individual components, and for cases of their serial and parallel connection. Then, for each calculation case the centroid of the corresponding density function is generated with subsequent plotting and comparison of the average graphs. Results. The primary observation based on the results of the graph analysis is that the average centroid resulting from two partial centroids of the density functions of single systems (mechanics, electronics, software) has a high rate of correlation (over 0.99) and almost matches the average centroid generated out of two partial centroids of serial and parallel connection of the respective pairs of systems per each calculation case. Conclusions. The results of the research again show that the average centroid for a composition of density functions of different systems is equivalent to their superposition and can be used as a conditional average (or fuzzy) index of the overall level of dependability of highly integrated complex technical systems of which the structural and functional dependability model is difficult to represent with a set of serial and parallel connections and corresponding mathematical tools of multiplication of probabilities.
Read full abstract